Chemical Reactivity of Sigma Singlet Oxygen O2(1Σg+) - American

Some of the most reactive O2(1∆g) scavengers have been used as substrates to explore the chemical reactivity of O2(1Σg+) in CCl4. No chemical deact...
0 downloads 0 Views 181KB Size
5672

J. Phys. Chem. A 1997, 101, 5672-5677

Chemical Reactivity of Sigma Singlet Oxygen O2(1Σg+) Marcus Bodesheim and Reinhard Schmidt* Institut fu¨ r Physikalische und Theoretische Chemie, J. W. Goethe-UniVersita¨ t, Marie-Curie-Str. 11, D60439 Frankfurt am Main, Germany ReceiVed: April 10, 1997; In Final Form: June 2, 1997X

Some of the most reactive O2(1∆g) scavengers have been used as substrates to explore the chemical reactivity of O2(1Σg+) in CCl4. No chemical deactivation channel could be found. However, an additional physical deactivation by the substrate was observed, which competes with the collision-induced deactivation of O2(1Σg+) by the substrate and which also leads to the formation of O2(1∆g). The rate constant of this additional deactivation of O2(1Σg+) is very similar to the rate constant of the deactivation of O2(1∆g) by the substrate.

Introduction Singlet oxygen is known to be a very reactive chemical species in photooxygenation reactions.1-5 However, there are two electronically excited singlet states lying above the 3Σgtriplet ground state of O2, which both could be involved in the singlet oxygen photochemistry. Their excitation energies amount to 94 kJ mol-1 for the 1∆g state and 157 kJ mol-1 for the 1Σg+ state.6 Because of the differences in energy and in electron correlation, it was already early speculated that both singlet state molecules could have a graduated reactivity and even perhaps their own different chemistry.7-10 The πg2 open shell configuration of the O2 molecule leads to six different electronic substates, one for 1Σg+, two for 1∆g, and three for 3Σ -. Kasha and Brabham concluded from the analysis of the g corresponding wave functions that different mechanisms could be expected for the reactions of O2 in the three lowest electronic states.10 The 1Σg+ state tends to have both antibonding π electrons in plane, so that concerted two-point addition reactions could be visualized. The 3Σg- state, in contrast, tends to have each of the π electrons oriented in mutually perpendicular planes. Therefore, intermediates with single-point attachment might be anticipated. The 1∆g state, however, was expected to show both concerted two-point addition and single-point attachment intermediates.10 Now it is known that the early photochemical investigations for the differentiation of the reactions of O2(1∆g) and O2(1Σg+) had been carried out in solvents, in which the lifetime τΣ of O2(1Σg+) is much too short for an efficient competition of a bimolecular chemical reaction with the fast physical deactivation of O2(1Σg+) by the solvent.11,12 Such experiments are further complicated by the fact that O2(1∆g), which is the product of the physical deactivation of O2(1Σg+),13,14 reacts very efficiently with the chosen substrates, due to its several orders of magnitude larger lifetime τ∆. The deactivation of O2(1∆g) is strongly forbidden, resulting in very long lifetimes in solution of 3.2 µs (H2O) e τ∆ e 0.3 s (perfuorodecalin).15,16 Thus, delta singlet oxygen is really a metastable species, enabling its well-known chemistry, which consists of the “ene” reaction, the 2 + 2, and the 2 + 4 cycloaddition reactions.1-4 The collision-induced deactivation O2(1∆g) f O2(3Σg-) occurs by spin-forbidden electronical to vibrational (e-v) energy transfer to high-frequency stretching modes of terminal bonds of the deactivating solvent molecule.17 The corresponding rate constant increases roughly exponentially with the fundamental vibrational energy of the deactivating bond.18,19 The same collisional e-v deactivation mechanism X

Abstract published in AdVance ACS Abstracts, July 1, 1997.

S1089-5639(97)01253-X CCC: $14.00

is also operating for sigma singlet oxygen,20-22 converting O2(1Σg+) completely to O2(1∆g).14 However, since this deactivation process is spin-allowed and since only 63 kJ mol-1 have to be transferred, rapid collisional deactivation takes place, reducing the lifetime τΣ of O2(1Σg+) in solution to very small values, e.g. 130 ps in C6H6.23 A maximum value of τΣ was found in CCl4 with 130 ns.23,24 Thus, in contrast to O2(1∆g), O2(1Σg+) can only be considered as a metastable species in perchlorinated solvents. Only after the first time-resolved measurements of O2(1Σg+) had been published23 did the development of promising strategies for answering the question of the chemical behavior of sigma singlet oxygen become possible. The first well-founded investigation of the chemical reactivity of O2(1Σg+) in solution was recently performed by Scurlock et al.25 The idea was that an efficient reaction of a reactive substrate with O2(1Σg+) in CCl4 would prevent its complete transformation to O2(1∆g) by collisional deactivation and thus reduce the efficiency of O2(1∆g) formation. Furthermore, in this case the experimental rate constants kQΣ of O2(1Σg+) quenching by the substrate should be larger than the values of kCΣ, which can be calculated for the collision-induced deactivation of O2(1Σg+) by the substrate following the universal e-v energy transfer mechanism.20-23 O2(1Σg+) and O2(1∆g) produced by photosensitization were monitored directly by observing the respective emissions at 1926 nm (1Σg+) and 1275 nm (1∆g).26,27 The 1Σg+ f 1∆g fluorescence could not be detected time resolved. Therefore, the values of kQΣ had to be obtained in Stern-Volmer experiments using the literature value of τΣ ) 130 ns in CCl4. Information about the efficiency of O2(1∆g) production in the absence or in the presence of added substrates was obtained by extrapolating the comparatively slowly (τ∆ g 10 µs) decaying 1∆g f 3Σgphosphorescence back to time zero. Furan and, in order to reduce the competition by the unavoidable collision-induced deactivation, partially or completely chlorinated dienes or olefins had been chosen as substrates. However, no chemical reaction of O2(1Σg+) could be detected. The experimental value of kQΣ was for none of the investigated substrates significantly larger than the value of kCΣ calculated for the collisional deactivation. The efficiency of O2(1∆g) sensitization was in no case reduced, even if the amounts of substrate added were sufficient to quench most of O2(1Σg+). These results definitely exclude any chemical reaction of O2(1Σg+) with the investigated substrates, which successfully could compete with the substrate-induced collisional 1Σ + f 1∆ deactivation by the e-v energy transfer.25 g g Like Scurlock et al. we also assume that olefins or dienes could react with O2(1Σg+) in reactions similar to the “ene” or 2 © 1997 American Chemical Society

Chemical Reactivity of O2(1Σg+) + 4 cycloaddition reactions of O2(1∆g). Both types of reactions occur with O2(1∆g) most probably via exciplexes with more or less strong charge transfer (CT).4,28-31 The exciplex decays either by physical deactivation or by a chemical reaction,28-31 which is assumed to take place via a two-step process (large CT)30 or in a more concerted way (small CT).29 Following Kasha and Brabham,10 concerted reaction paths could be envisaged for the reaction of O2(1Σg+) with olefins and dienes. Due to its higher energy content, much larger rate constants could be expected than for O2(1∆g). Therefore, the possibility that some substrates, which are very reactive O2(1∆g) scavengers, might be sufficiently reactive toward O2(1Σg+) to surpass the physical deactivation by the e-v energy transfer is still not excluded. It was already mentioned by Scurlock et al. that the introduction of chlorine atoms as substituents of the substrates could lead to a reduced reactivity with singlet oxygen25 because of the concomittant reduction of π-electron density. Furthermore, furan, which presumably was the most reactive substrate toward O2(1∆g) in the study of Scurlock et al., deactivates O2(1∆g) with a rate constant of only kQ∆ ) 3.6 × 106 M-1 s-1.32 There are some substrates mentioned in the literature5 which react much faster with O2(1∆g) and which could be better candidates for the discovery or the exclusion of sigma singlet oxygen chemistry. We chose substrates with maximum rate constants of kQ∆ ) 6 × 108 M-1 s-1 in CCl4. If these compounds, for which kQ∆ is not much smaller than the rate constant kCΣ of collisional deactivation of O2(1Σg+), are added in sufficient amounts to quench most of O2(1Σg+), the lifetime of O2(1∆g) is reduced to a few hundred nanoseconds. Thus, an accurate determination of the effect of the substrate on the efficiency of O2(1∆g) formation is no more possible by extrapolation of the phosphorescence intensity to time zero. We obtained information about the efficiency of O2(1∆g) formation in the presence of a substrate Q by comparing the ratios of the integrated 1∆g emission intensities over the corresponding 1∆g lifetimes of solutions with added substrate Q and with added collisional quencher benzene. Furthermore, we used a different technique, which allowed us to monitor O2(1Σg+) time resolved by observing the 1Σg+ f 3Σg- emission at 765 nm.23 Since the O2(1Σg+) lifetime in CCl4 is very sensitive to contaminations by the solvent, the time-resolved measurements enable a more accurate determination of the rate constant kQΣ compared with the evaluation of Stern-Volmer experiments basing on 1Σg+ f 1∆g emission intensities and the literature value of τΣ. For the most reactive compounds we found rate constants kQΣ which significantly surpass the calculated values of kCΣ. Thus, our study complements and extends the previous work on the reactivity of sigma singlet oxygen.25 Experimental Details CCl4 (TET, Janssen, 99+%), and the Aldrich chemicals benzene (BNZ, 99+%), 2,3-dimethyl-2-butene (tetramethylethylene, TME, 99+%), furan (99+%), 2-methylfuran (MEF, 99%), 2-methoxyfuran (MOF, 97%), 2,5-dimethylfuran (DMF, 99%), pyrrole (PYR, 99%), N-(2-cyanethyl)pyrrole (NCP, 99+%), N-methylpyrrole (NMP, 99%), and phenalenone (PHE, Aldrich, 97%) were purified by column chromatography using Al2O3 or silica gel (PHE). Phenazine (PHZ, Fluka > 98%) and 7H-benz[de]anthracen-7-one (benzanthrone, BAN, Aldrich) were recrystallized from ethanol. Solutions were prepared and filled into sample cells in a glovebox under dry atmosphere. The principal setup for the time-resolved measurements of the O2(1Σg+) and O2(1∆g) phosphorescences has been described.14,23,33 Briefly, we used as excitation source an excimer laser filled with N2 (Lambda Physik, EMG 101 E, 337 nm, pulse width 7

J. Phys. Chem. A, Vol. 101, No. 31, 1997 5673 ns). Alternatively a dye laser FL 3002, which was pumped by an EMG 200 E excimer laser (both from Lambda Physik) was used for excitation at 400 nm. The sample housing allows the simultaneous observation of two emissions in T arrangement. In this way, the formation and decay of O2(1Σg+) and O2(1∆g) can be followed simultaneously by their corresponding phosphorescence traces at 765 nm (R1464 photomultiplier, Hamamatsu) and 1276 (liquid N2 cooled germanium diode EO 817P, North Coast). The emissions were isolated by suited interference filters. The purified quenchers did not absorb at 337 nm with the exception of MOF, which was excited at 400 nm. The absorbances of the sensitizers amounted to about 1 per cm at the excitation wavelengths and were kept constant in the quenching experiments. The time-resolved experiments allowed control of the triplet state lifetime τT of the sensitizers. τT was not affected by the addition of the quenchers used. All emission experiments were done with air-saturated solutions at room temperature varying the laser pulse energy. Only energyindependent results are reported. Photochemical experiments for the determination of the ratio kR∆/kQ∆ of the rate constants of chemical and overall deactivation of O2(1∆g) by NMP were performed under continuous irradiation of the sensitizers PHE or PHZ in a setup described already earlier.34 The oxygen consumption rate was measured in comparison to the scavenger TME, which is known to deactivate O2(1∆g) exclusively by chemical reaction.29,35 Results and Discussion 1Σ ≡ O (1Σ +), 1∆ ≡ O (1∆ ), and 3Σ ≡ O (3Σ -) are formed 2 g 2 g 2 g with corresponding efficiencies a, b, and 1 - a - b in the photosensitization of O2 by the triplet state (T1) sensitizers PHE, PHZ and BAN in air-saturated TET.36 The radiative deactivation of T1, 1Σ, and 1∆ is negligible.14,23,33 The following equations describe the principal deactivation paths of the different excited species.

T1 + 3Σ f S0 + 1Σ

akOT

(1)

T1 + 3Σ f S0 + 1∆

bkOT

(2)

T1 + 3Σ f S0 + 3Σ

(1 - a - b)kOT

(3)

1

kMΣ

(4)

1

kCΣ

(5)

1

kRΣ

(6)

1

kM∆

(7)

1

kP∆

(8)

1

kR∆

(9)

Σ + M f M + 1∆ Σ + Q f Q + 1∆ Σ + Q f PΣ ∆ + M f M + 3Σ ∆ + Q f Q + 3Σ ∆ + Q f P∆

S0, M, and Q represent the ground state sensitizer, solvent molecules and the added substrates listed in Table 1. PΣ and P∆ are the products of reactions of 1Σ and 1∆ with Q. The superscripts T, Σ, and ∆ of the rate constants k denote the excited state, which is quenched. The subscripts O and M refer to quencher O2 and solvent M. The subscripts P and R indicate the physical and the reactive deactivation by the added substrate Q. Thus, kQ∆ ) kP∆ + kR∆ holds true. kCΣ is the rate constant of collisional deactivation of O2(1Σg+) by Q following the e-v energy transfer. We obtain as lifetimes of T1, 1Σ, and 1∆:

5674 J. Phys. Chem. A, Vol. 101, No. 31, 1997

Bodesheim and Schmidt

TABLE 1: Experimental Rate Constants kQ∆ and kQΣ of Quenching of O2(1∆g) and O2(1Σg+) by Different Substrates Q, and Rate Constants kCΣ of Collisional Quenching of O2(1Σg+) Calculated According to the e-v Energy Transfer Mechanisma

c

Q

k Q∆

kQΣ

kCΣ

kCΣ + kQ∆

TMEb MEFb MOFb DMFb PYRd NCPc NMPd

0.15 ( 0.03 0.6 ( 0.1 1.5 ( 0.1 3.2 ( 0.4 1.2 ( 0.1 3.1 ( 0.7 6.0 ( 0.5

11 ( 1.7 7.8 ( 1.5 6.3 ( 1.1 9.8 ( 1.5 12.1 ( 2.9 11.6 ( 2.7 13.5 ( 1.4

12 ( 2 6(1 6(1 8 ( 1.5 9 ( 1.5 8 ( 1.5 7(1

12.2 ( 2 6.6 ( 1 7.5 ( 1 11.2 ( 1.5 10.2 ( 1.5 11.1 ( 1.5 13.0 ( 1

a All rate constants are in units of 108 M-1 s-1. b Sensitizer PHE. Sensitizers PHE and PHZ. d Sensitizers PHE, PHZ, and BAN.

Figure 2. Correlation of 1/τ∆ with [NMP] in TET + 7 vol % BNZ. kQ∆ ) (6.3 ( 0.2) × 108 M-1 s-1. Inset: Rise and decay of the O2(1∆g) phosphorescence at 1275 nm. Open circles are experimental data, curve represents convolution of apparatus function with eq 12, τT ) 245 ns and τ∆ ) 353 ns. Sensitizer PHZ, [NMP] ) 0.004 63 M.

apparatus function (Ge diode response to the laser pulse) to the experimental 1∆ phosphorescence curves I1275(t). However, if 7 vol % BNZ are added to the solution, 1Σ is very rapidly quenched by collisional deactivation and c ) 1, τΣ ) 1.8 ns , τ∆, τT results, leading to eq 12. Then, of course, the lifetime of 1∆ is reduced to about 350 µs already in the absence of substrate Q. Figure 1. Correlation of 1/τΣ with [NMP] in TET. kQΣ ) (14.7 ( 1.0) × 108 M-1 s-1.

τT ) (kOT[O2])-1, τΣ ) (kMΣ + (kCΣ + kRΣ)[Q])-1 and τ∆ ) (kM∆ + (kP∆ + kR∆)[Q])-1. Equation 10 describes the rise and decay of [1Σ]t. The

[1Σ]t ) [T1]0aτΣ/(τΣ - τT){exp(-t/τΣ) - exp(-t/τT)} (10) emission I765(t) observed at 765 nm consists of a very intense, fast-decaying background luminescence L(t) followed by the 1Σ phosphorescence.14,23 If 7 vol % BNZ are added to the solution, the 1Σ emission is quenched completely, whereas L(t) remains unchanged. Therefore, the difference signal ∆I765(t) ) I765(t) - L(t) is directly proportional to [1Σ]t. Lifetimes τT and τΣ are obtained by nonlinear least-squares fitting of convolutions of the apparatus function (photomultiplier response to the laser pulse) with eq 10 to the experimental curves ∆I765(t).23 Variation of [Q] influenced τΣ but not τT, which demonstrates that the substrate Q does not deactivate the sensitizer triplet. Plots of 1/τΣ versus [Q] are linear as is exemplarily demonstrated in Figure 1 for the substrate NMP. The slope of the linear least-squares fit is the overall quenching constant kQΣ ) kCΣ + kRΣ. Values of kQΣ obtained for the different substrates are listed in Table 1. The rise and decay of [1∆]t follows eq 11 with A ) [T1]0τ∆/

[1∆]t ) A{exp(-t/τ∆) - exp(-t/τT)} + B{exp(-t/τΣ) - exp(-t/τT)} (11) (τ∆ - τT){b + acτ∆/(τ∆ - τΣ)}, B ) [T1]0acτΣτ∆/(τ∆ - τΣ)/(τT - τΣ), and c being the efficiency of 1∆ formation in the 1Σ deactivation. If 1Σ is only quenched by collisional deactivation according to the e-v mechanism as it is the case for TET and BNZ, then c ) 1.14 Equation 11 is too complex to be used in the nonlinear least-squares fitting of convolutions with the

[1∆]t,B ) [T1]0(b + a)τ∆/(τ∆ - τT){exp(-t/τ∆) exp(-t/τT)} (12) Equation 12 is used to fit the 1∆ phosphorescence I1275,B(t) in mixtures of TET and 7 vol % BNZ by convolution techniques. τ∆ and τT are obtained. If the substrate Q is present in the solvent mixture in such high concentrations that deactivation of 1∆ occurs only by Q (τ∆ ≈ 1 µs), then τ∆ is reciprocally related with [Q]. The value of τT was not affected by the substrates. Figure 2 demonstrates exemplarily for the substrate NMP the linear correlation of 1/τ∆ with [Q]. The slope of the linear least-squares fit is the overall quenching constant kQ∆ ) kP∆ + kR∆. Values of kQ∆ obtained for the different substrates are collected in Table 1. Our values of kQ∆ are in line with literature data. kQ∆ depends only moderately on the solvent polarity for TME, which is a purely chemical quencher of O2(1∆g).35 For example, kQ∆ amounts (in units of 108 M-1 s-1) to 0.64 (benzonitrile),37 0.42 (CH2Cl2),37 0.32, 0.36 (toluene),37,38 0.22 (CS2),39 and 0.092 (n-hexane).37 The furan derivatives deactivate O2(1∆g) also only chemically.29 Values of 1.0 (CH3OH)29 and 0.62 (CH2Cl2)40 have been reported for MEF, 1.6 (CH3OH),29 and 1.2 (CH2Cl2)40 for MOF, 1.8 (CH3OH),29 1.6 (CH2Cl2),40 and 6.3 (toluene)41 for DMF. Less data are available for pyrrole derivatives: 1.7 (toluene)42 for PYR, for which we obtained 2.4 (toluene) and 4.7 (CH3CN), and 3.7 (toluene)42 and 10 (CCl4)43 for NMP. Rate constants of collisional deactivation of singlet oxygen are composed additively from incremental rate constants kXY of deactivation by terminal bonds X-Y. This has been shown to hold true for O2(1∆g)17-20 as well as for O2(1Σg+).20-23 From our data on the collisional deactivation of O2(1Σg+) we calculate kCHΣ ) (1.0 ( 0.2) × 108 23 and kNHΣ ) (5 ( 1.5) × 108 M-1 s-1.44 Both numbers can be taken to calculate the rate constants kCΣ of collisional deactivation of O2(1Σg+) according to the e-v energy transfer for the substrates used. These data are given for means of comparison in Table 1. For TME we find kQΣ

Chemical Reactivity of O2(1Σg+)

J. Phys. Chem. A, Vol. 101, No. 31, 1997 5675 TABLE 2: Comparison of the Substrate Concentration Dependence of Experimental and Calculated Values of R ) (b + acQ)/(b + a)a [PYR], M

Rexp

Rcalc

[NMP], M

Rexp

Rcalc

0 0.0037 0.0074 0.0111 0.0148 0.0185

1.00 ( 0.05 1.12 ( 0.09 1.00 ( 0.06 0.96 ( 0.06 1.03 ( 0.13 1.01 ( 0.15

1.00 0.92 0.88 0.86 0.85 0.84

0 0.0024 0.0048 0.0072 0.0096 0.0120

1.00 ( 0.05 1.05 ( 0.08 1.04 ( 0.08 1.05 ( 0.08 1.06 ( 0.13 1.02 ( 0.18

1.00 0.88 0.81 0.77 0.74 0.72

a

Details are described in the text.

represents the overall efficiency of indirectly (ac) via O2(1Σg+) and directly (b) formed O2(1∆g) in the sensitization of O2 by T1. Equation 13 allows the investigation, on whether a substrate Figure 3. Correlation of kQΣ - kCΣ with kQ∆ data of Table 1. The straight line represents the linear least-squares fit with slope 1.0 ( 0.3 and intercept 0.0 ( 1.5.

slightly smaller than kCΣ, but considering the error limits kQΣ ) kCΣ. Obviously, kRΣ must be much smaller than kCΣ. Therefore, the photooxygenation of TME takes place only via O2(1∆g). The possibility of the reaction of O2(1Σg+) with TME and 1,1dimethyl-2,2-dimethoxyethylene was already earlier investigated in gas phase experiments by Hammond.45 O2(1Σg+) was produced in mixtures of O2 and olefin by resonant absorption 1Σ + r 3Σ -. His findings allowed only two alternative g g conclusions: (1) The olefins react with both O2(1Σg+) and O2(1∆g) under formation of identical products or (2) O2(1Σg+) is collisionally deactivated to O2(1∆g), before the reaction of olefin and O2(1∆g) takes place. Our results confirm Hammond’s conclusion (2). Since TME and 1,1-dimethyl-2,2-dimethoxyethylene belong to the most reactive olefins, it appears that O2(1Σg+) is chemically not reactive with olefins. For MEF, MOF, and DMF, which are more reactive O2(1∆g) scavengers, kQΣ is larger than kCΣ, but the deviation is only about as large as the experimental uncertainty. However, for PYR, NCP, and NMP kQΣ is distinctly larger than kCΣ. If this finding would have been made only for one of the pyrroles, we would deny the significance of the difference, which actually is only a little bit larger than the error limits. But as we find a distinct difference for all three pyrroles, and since there is a trend for increasing difference kQΣ - kCΣ with increasing value of kQ∆, which is illustrated by the correlation of Figure 3, we take the inequality kQΣ > kCΣ to be true. Thus, for the first time besides the ubiquitous collisioninduced deactivation an additional channel for the deactivation of O2(1Σg+) is observed, which could well be a chemical reaction with the substrates. Table 1 compares the sums kCΣ + kQ∆ with the experimental rate constants kQΣ. A good agreement between these values has to be noted. This finding demonstrates that O2(1Σg+) reacts on no account faster with the investigated substrates than O2(1∆g). This is a surprising result, since one would expect a much higher reactivity of O2(1Σg+) because of its by 63 kJ mol-1 larger excitation energy. The correlation of Figure 3 of kQΣ - kCΣ versus kQ∆ with slope 1.0 ( 0.3 shows that the additional deactivation of O2(1Σg+) by Q occurs with approximately the same rate constant as the deactivation of O2(1∆g) by Q. This very interesting result will be discussed below. If the additional deactivation channel for O2(1Σg+) is actually due to a chemical reaction with the substrate Q, the efficiency of the photosensitized O2(1∆g) formation must be reduced in the presence of Q. Integration of [1∆]t (eq 11) in the limits t ) 0 to ∞ leads to the simple eq 13, where the sum b + ac

INT∆ ) [T1]0τ∆(b + ac)

(13)

Q reduces the efficiency of 1∆ formation in 1Σ deactivation. In the absence of BNZ and in the presence of a 1Σ deactivating substrate Q the efficiency of 1∆ formation in 1Σ deactivation is cQ. If Q deactivates 1Σ also by chemical reaction, cQ becomes smaller than unity. The ratio of the integrated 1∆ phosphorescence intensity INT1275 over the respective 1∆ lifetime is then given by eq 14, where f is a constant which is directly proportional to the rate constant of the 1∆g f 3Σg- emission. If

(INT1275/τ∆)Q ) f[T1]0(b + acQ)

(14)

no substrate Q but 7 vol % BNZ are present in the TET solution of the sensitizer c ) 1, and the respective ratio (INT1275/τ∆)B is given by eq 15 with constant f′.

(INT1275/τ∆)B ) f′[T1]0(b + a)

(15)

In these experiments we have to consider that the radiative rate constant of the 1∆g f 3Σg- emission depends on the solvent.46 The relative rate constant of emission amounts to 0.73 s-1 in TET related to 1 s-1 in BNZ.33 It was shown by us that the emission is a bimolecular collision-induced process.33 Therefore, we calculate the relative values of the second-order rate constants of phosphorescence of 0.071 M-1 s-1 (TET) and 0.089 M-1 s-1 (BNZ) and from the molarities of TET and BNZ in the mixture of 7 vol % BNZ in TET f′ ) 1.025f. Experiments were performed with BAN, which sensitizes O2(1Σg+) and O2(1∆g) with efficiencies of a ) 0.85 and b ) 0.15, respectively.36 NMP and PYR were the substrates Q. We determined the experimental energy normalized ratios Rexp ) 1.025(INT1275/ τ∆)Q/(INT1275/τ∆)B ) (b + acQ)/(b + a), which are listed in Table 2. Efficiencies cQ can be estimated by eq 16, kMΣ ) (130 ns)-1,

cQ ) (kMΣ + kCΣ[Q])/(kMΣ + kQΣ[Q])

(16)

and the data of kQΣ and kCΣ. These values can then be used to calculate ratios Rcalc ) (b + acQ)/(b + a), which are compared in Table 2 with values of Rexp. R represents the ratio of the overall efficiency of O2(1∆g) formation in the presence of a substrate in concentration [Q] over the overall efficiency of O2(1∆g) formation under conditions of pure collision induced deactivation of O2(1Σg+). For PYR a decrease of Rcalc by 16% is expected at the maximum substrate concentration if Q would react with O2(1Σg+) with rate constant kQΣ - kCΣ. But we observe no significant decrease of Rexp with increasing [PYR]. The data scatter around a mean value of 1.00, if the deviating value Rexp ) 1.12 at [PYR] ) 0.037 M is neglected. The results

5676 J. Phys. Chem. A, Vol. 101, No. 31, 1997

Bodesheim and Schmidt

are even more definite for NMP, where a decrease of Rcalc by 28% is expected but no decrease of Rexp is found. These findings show that the additional O2(1Σg+) deactivation channel leads to the formation of O2(1∆g). There is neither a particular chemical reaction of O2(1Σg+) nor a physical process which bypasses O2(1∆g) in the additional deactivation of O2(1Σg+). The net effect of the additional physical deactivation process is quantitative conversion 1Σ f 1∆. Thus eq 6 of the reaction scheme has to be replaced by eq 6a with rate constant 1

Σ + Q f Q + 1∆

kPΣ

(6a)

kPΣ ) kQΣ - kCΣ. The observation kPΣ ≈ kQ∆ (compare Figure 3) lets us suspect that the deactivation process 6a is intimately related with the deactivation of O2(1∆g) via the processes 8 and 9. The reaction of O2(1∆g) with cyclic dienes occurs as 2 + 4 cycloaddition forming endoperoxides in the primary step.1-4 In temperature-dependent investigations (olefins, enol ethers, furan derivatives, indoles), it was found that the reaction requires practically no activation enthalpy.38,39,41,47 Therefore, it is assumed that an exciplex with some CT is formed as intermediate.4,28,38 Generally, for most substrates a competition between physicalspossibly CT mediated48,49sand chemical deactivation of O2(1∆g) is observed, whereby already small sterical variations of the substrate can be sufficient to change the balance between the deactivation processes drastically.50 There is strong evidence that the competing processes occur via the same intermediate, which is formed in the rate-determining step,28,30,31 and which collapses either under formation of an endoperoxide or in the physical deactivation step. Whereas it is known that the deactivation of O2(1∆g) by MEF, MOF, and DMF occurs completely by chemical reaction, i.e., kR∆ ) kQ∆,29 no respective data have been found for the pyrrole derivatives, which are also known to react with O2(1∆g).2 We therefore determined under identical sensitization conditions the O2-consumption rate by NMP in comparison to the scavenger TME. The experiments resulted in kR∆ e 0.1kQ∆, i.e., O2(1∆g) is deactivated by NMP principally physically, probably by a CT mediated process. However, since the chemical reactions of pyrroles with O2(1∆g) proceed also via endoperoxides,2 similar intermediates as for the furan derivatives can be assumed. A transition state in which the oxygen atoms lie above the carbon atoms in the 2- and 5-positions of the heterocycle seems reasonable. Since ∆Hq ≈ 0, the activation entropy ∆Sq determines the value of kQ∆ for the chemical and/or physical deactivation of O2(1∆g). ∆Hq ≈ 0 can also be expected for the deactivation of O2(1Σg+), because of its larger excitation energy. If one accepts this assumption, the finding kPΣ ≈ kQ∆ demonstrates that the O2(1Σg+) deactivation and the deactivation of O2(1∆g) require practically the same activation entropy, indicating the same transition states. Therefore, we propose that the deactivation of O2(1Σg+) by process 6a proceeds via the transition state for the deactivation of O2(1∆g) by the heterocyclic diene: As soon as the new bonds between oxygen and carbon atoms begin to form in the approach to the transition state, the door to a large heat bath is opened, in which the excess excitation energy of O2(1Σg+) versus O2(1∆g) is instantaneously released as heat, leading to dissociation into O2(1∆g) and substrate. 63 kJ mol-1 have to be distributed. Assuming that the excess energy is completely transferred to the heat bath of the substrate, we estimate from the molar heat capacity of Cp ) 128 J mol-1 K-1 of PYR51 a maximum value for the instantaneous jump of the internal temperature of PYR of about 500 K. If the energy is distributed in the way that both PYR and O2(1∆g) (Cp ) 29 J mol-1 K-1)51 take the same temperature, then the temperature jump still amounts to about 400 K, which certainly is sufficient

to explain the observed dissociation into O2(1∆g) and substrate subsequent to deactivation. Furthermore, dissociation is preferred to reaction of O2(1∆g) and substrate as the values of kQ∆ are at least about 2 orders of magnitude smaller than the rate constant of diffusion, indicating a very small reaction probability of O2(1∆g) and substrate per encounter. The deactivation process 6a is electronically allowed like the collision-induced O2(1Σg+) f O2(1∆g) transition, for which the electronic factor is near by unity.22 In addition, the coupling between the heat bath of the substrate and O2(1Σg+) should become much stronger during the approach to the transition state than in a simple collision, where no bonds are formed. The proposed coupling of O2(1Σg+) with the heat bath of the substrate in the region of the transition state can, however, not induce the spin-forbidden deactivation process 1Σ f 3Σ because of the corresponding poor electronic factor. For example, the ratio of the electronic factors of the collision-induced deactivation processes 1∆ f 3Σ and 1Σ f 1∆ was determined to be only 6 × 10-5.20,22 The lack of formation of an endoperoxide PO by a chemical reaction of O2(1Σg+) with Q can be understood by a consideration of the corresponding energy balance. POs of aromatic compounds split thermally with activation enthalpies between 125 and 140 kJ mol-1 into O2(1∆g) and aromatic substrate.52-56 Assuming ∆Hq for the reverse reaction to be close to zero, results in an estimate for the reaction enthalpy of PO formation from ground state O2 and aromatic compound of ∆H ≈ -30 to -45 kJ mol-1. Therefore, the one-step formation of PO from Q and O2(1Σg+) would release an amount of energy of roughly 190 kJ mol-1, which is much larger than the activation enthalpy of PO cleavage and would provoke dissociation into O2(1∆g) and Q. Actually, it was already shown for the PO of 2,5diphenylfuran that the thermal cleavage of the PO yields furan and O2(1∆g).57,58 Thus, the immediate dissociation after deactivation 1Σ f 1∆ in the region of the transition state of PO formation appears to be reasonable. Conclusions Our investigations on the chemical reactivity O2(1Σg+) were focused on some of the most reactive O2(1∆g) scavengers. No chemical deactivation channel of O2(1Σg+) could be detected. However, we observed an additional physical deactivation process, which competes with the general collision-induced deactivation of O2(1Σg+) by the e-v energy transfer and which also leads to the formation of O2(1∆g). Since kPΣ is very similar to kQ∆, it is assumed that O2(1Σg+) walks on the reaction path of O2(1∆g) approaching the same transition state, at least for the furan and pyrrole derivatives. It is assumed that deactivation 1Σ f 1∆ occurs in the region of the transition state of PO formation. As soon as the new bonds between oxygen and carbon atoms begin to form, the door to a large heat bath is opened, in which the excess excitation energy of O2(1Σg+) versus O2(1∆g) is instantaneously released as heat leading to dissociation into O2(1∆g) and substrate. Obviously, the chemistry of O2(1Σg+) is prevented by its lability with respect to spin-allowed deactivation. The approach to any transition state would lead with beginning bond formation to an efficient coupling of the heat bath of the substrate to O2(1Σg+) inducing instantaneously deactivation and formation of O2(1∆g). Therefore, we conclude that O2(1Σg+) is generally chemically not reactive. Acknowledgment. We thank Prof. C. Tanielian, EHICS, Strasbourg, for the opportunity to do the photochemical experiments in his laboratory and Dr. R. Mechin for kind assistance. Financial support by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie is gratefully acknowledged.

Chemical Reactivity of O2(1Σg+) References and Notes (1) Ranby, B., Rabek, J. F., Eds. Singlet Oxygen; Wiley & Sons: New York, 1978. (2) Wasserman, H. H.; Lipshutz, B. H. In Singlet Oxygen; Wasserman, H. H., Murray, R. W., Eds.; Academic Press: New York, 1979; p 430. (3) Frimer, A. A., Ed. Singlet Oxygen; CRC Press: Boca Raton, FL, 1985; Vols. I-IV. (4) Gorman, A. A. AdV. Photochem. 1992, 17, 217. (5) Wilkinson, F.; Helman, W. P.; Ross, A. B. J. Phys. Chem. Ref. Data 1995, 24, 663. (6) Herzberg, G. Spectra of Diatomic Molecules; Van Nostrand Reinhold: New York, 1950. (7) Kawaoka, K.; Khan, A. U.; Kearns, D. R. J. Chem. Phys. 1967, 46, 1842. (8) Kawaoka, K.; Khan, A. U.; Kearns, D. R. J. Chem. Phys. 1967, 47, 1883. (9) Kearns, D. R.; Khan, A. U. Photochem. Photobiol. 1969, 10, 193. (10) Kasha, M.; Brabham, D. E. In Singlet Oxygen; Wasserman, H. H., Murray, R. W., Eds.; Academic Press: New York, 1979; p 1. (11) Kearns, D. R.; Hollins, R. A.; Khan, A. U.; Chambers, R. W.; Radlick, P. J. Am. Chem. Soc. 1967, 89, 5455. (12) Kearns, D. R.; Hollins, R. A.; Khan, A. U.; Radlick, P. J. Am. Chem. Soc. 1967, 89, 5456. (13) Wildt, J.; Bednarek, G.; Fink, E. H.; Wayne, R. P. Chem. Phys. 1988, 122, 1842. (14) Schmidt, R.; Bodesheim, M. Chem. Phys. Lett. 1993, 213, 111. (15) Egorov, S. Y.; Kamalov, V. F.; Koroteev, N. I.; Krasnovskii, A. A.; Toleutaev, B. N.; Zinukov, S. V. Chem. Phys. Lett. 1989, 163, 421. (16) Afshari, E.; Schmidt, R. Chem. Phys. Lett. 1991, 184, 128. (17) Hurst, J. R.; Schuster, G. B. J. Am. Chem. Soc. 1983, 105, 5756. (18) Schmidt, R.; Brauer, H.-D. J. Am. Chem. Soc. 1987, 109, 6976. (19) Schmidt, R.; Afshari, E. Ber. Bunsenges. Phys. Chem. 1992, 96, 788. (20) Schmidt, R. Ber. Bunsenges. Phys. Chem. 1992, 96, 794. (21) Wang, B.; Ogilby, P. R. J. Phys. Chem. 1993, 97, 193. (22) Schmidt, R. J. Photochem. Photobiol. A: Chem. 1994, 80, 1. (23) Schmidt, R.; Bodesheim, M. J. Phys. Chem. 1994, 98, 2874. (24) Chou, P.-T.; Wei, G.-T.; Lin, C.-H.; Wei, C.-Y.; Chang, C.-H. J. Am. Chem. Soc. 1996, 118, 3031. (25) Scurlock, R. D.; Wang, B.; Ogilby, P. R. J. Am. Chem. Soc. 1996, 118, 388. (26) Chou, P. T.; Frei, H. Chem. Phys. Lett. 1985, 122, 87. (27) Wang, B.; Ogilby, P. R. J. Phys. Chem. 1993, 97, 9593. (28) Gorman, A. A.; Gould, I. R.; Hamblett, I.; Standen, M. L. J. Am. Chem. Soc. 1984, 106, 6956. (29) Gollnick, K.; Griesbeck, A. Tetrahedron 1985, 41, 2057. (30) Aubry, J.-M.; Manard-Cazin, B.; Rougee, M.; Bensasson, R. V. J. Am. Chem. Soc. 1995, 117, 9159.

J. Phys. Chem. A, Vol. 101, No. 31, 1997 5677 (31) Gollnick, K.; Lindner, J. H. E. Tetrahedron Lett. 1973, 1903. (32) Our value disagrees with kQ∆ ) 9.0 × 106 M-1 s-1 given by Scurlock et al. for furan in CCl4.25 The reason for this discrepancy is not known. Literature data: 9 × 106 (CH3OH),29 1.4 × 107 (CD3OD),30 4.5 × 107 M-1 s-1 (D2O).30 (33) Schmidt, R.; Bodesheim, M. J. Phys. Chem. 1995, 99, 15919. (34) Schmidt, R.; Tanielian, C.; Dunsbach, R.; Wolff, C. J. Photochem. Photobiol. A: Chem. 1994, 79, 11. (35) Griesbeck, A.; Gollnick, K. Tetrahedron Lett. 1984, 25, 725. (36) Bodesheim, M.; Schu¨tz, M.; Schmidt, R. Chem. Phys. Lett. 1994, 221, 7. (37) Hild, M.; Brauer, H.-D. Ber. Bunsenges. Phys. Chem. 1996, 100, 1814. (38) Gorman, A. A.; Hamblett, I.; Lambert, C.; Spencer, B.; Standen, M. L. J. Am. Chem. Soc. 1988, 110, 8053. (39) Hurst, J. R.; Schuster, G. B. J. Am. Chem. Soc. 1982, 104, 6854. (40) Clennan, E. L.; Mehrsheikh-Mohammadi, M. E. J. Am. Chem. Soc. 1984, 106, 7112. (41) Gorman, A. A.; Lovering, G.; Rodgers, M. A. J. J. Am. Chem. Soc. 1979, 101, 3050. (42) Venediktov, E. A.; Titova, Y. V. Biofizika 1993, 38, 1031. (43) Venediktov, E. A.; Perfil’ev, V. A.; Titova, Y. V.; Karavaev, B. I.; Berezin, B. D. Biofizika 1987, 32, 203. (44) To be published separately. (45) Hammond, W. B. Tetrahedron Lett. 1979, 2309. (46) Scurlock, R. D.; Ogilby, P. R. J. Phys. Chem. 1987, 91, 4599. (47) Gorman, A. A.; Gould, I. R.; Hamblett, I. J. Am. Chem. Soc. 1982, 104, 7098. (48) Ogryzlo, E. A.; Tang, C. W. J. Am. Chem. Soc. 1970, 92, 5034. (49) Young, R. H.; Martin, R. L.; Feriozi, D.; Brewer, D.; Kayser, R. Photochem. Photobiol. 1973, 17, 233. (50) Acs, A.; Schmidt, R.; Brauer, H.-D. Ber. Bunsenges. Phys. Chem. 1987, 91, 1331. (51) Lide, D. R., Ed. Handbook of Chemistry and Physics, 74th ed.; CRC Press: Boca Raton, FL, 1993-1994; pp 6-131. (52) Wassermann, H. H.; Scheffer, J. R.; Cooper, L. J. Am. Chem. Soc. 1972, 94, 4991. (53) Turro, N. J.; Chouw, M.-F.; Rigaudy, J. J. Am. Chem. Soc. 1979, 101, 1300. (54) Turro, N. J.; Chouw, M.-F. J. Am. Chem. Soc. 1979, 101, 3701. (55) Schmidt, R.; Drews, W.; Brauer, H.-D. J. Am. Chem. Soc. 1980, 102, 2791. (56) Brauer, H.-D.; Schmidt, R. In Photochromism; Du¨rr, H., BouasLaurent, H., Eds.; Elsevier: Amsterdam, 1990. (57) Dufraisse, C.; Rio, G.; Ranjon, A. C. R. Acad. Sci. Paris 1967, 264, 516. (58) Trozollo, A. M.; Fahrenholtz, S. R. Ann. N. Y. Acad. Sci. 1970, 171, 61.